=== chriadam|away is now known as chriadam === salem_ is now known as _salem === chihchun_afk is now known as chihchun === vying is now known as Guest54115 === chihchun is now known as chihchun_afk === chihchun_afk is now known as chihchun === cul_ is now known as cul === jackson is now known as Guest50945 [06:07] so no more U1 on phone either? That's a shame :( Are there yet any plans to support other cloud services? [06:09] Owncloud support would be rather nice alternative along dropbox and google drive. One Drive can be omitted :D [06:11] Hm. Is /dev/input really meant to be readable by the phablet user? [06:54] good morning === LarrySteeze is now known as LarrySteeze|Away [08:01] cjwatson: ping [08:01] cjwatson: I tried to backport the 0.4.20 and 0.4.21 to 13.10 but the tests failed: https://launchpadlibrarian.net/171659457/buildlog_ubuntu-saucy-amd64.click_0.4.20ubuntu1_FAILEDTOBUILD.txt.gz === chihchun is now known as chihchun_afk [08:08] Hm, I’m on r250, current image is r274, but Ubuntu says “Software is up to date” [08:08] Is there a dashboard somewhere I can see which is the latest promoted image? [08:13] mpt, nope, you can fish that info out of a json file though (read: someone could write such a dashboard) http://system-image.ubuntu.com/ubuntu-touch/trusty/mako/index.json (last line) [08:14] Huh, so r250 really is the latest promoted, ok === chriadam is now known as chriadam|away [08:15] Thanks ogra_ [08:17] Good morning all; happy Find a Rainbow Day! :-D [08:26] bzoltan: OK, I'll have a look [08:26] cjwatson: thank you [08:52] bzoltan: http://bazaar.launchpad.net/~click-hackers/click/trunk/revision/412 === chihchun_afk is now known as chihchun [09:03] cjwatson: thanks === vrruiz_ is now known as rvr [10:11] Morning all === chihchun is now known as chihchun_afk === chihchun_afk is now known as chihchun === MacSlow is now known as MacSlow|lunch === dandrader is now known as dandrader|afk [10:37] ogra_, any idea what sets the HOSTNAME env var to "android" on touch, and if it’s a reliable way of detecting we’re running on touch? [10:39] no it is not ... thats a hack from before the container flip [10:39] iirc it sits in /etc/environment ... and we will likely drop it in some cleanup session [10:39] ok [10:40] also touch will not necessarily use an android container in the future (at some point at least) [10:42] oSoMoN, congrats btw, i'm just looking at http://oos.moxiecode.com/js_webgl/fur/ on my phone ... apart from some flickering it works really nicely [10:44] ogra_, kudos go to chrisccoulson, he’s the oxide mastermind :) [10:44] ++ === dandrader|afk is now known as dandrader [10:47] * chrisccoulson blushes === vying is now known as Guest5733 [11:18] hy guys, could anybody guide me with some ubuntu touch install questions, i'm stack for the last few days installing it, after complete install it will not boot [11:19] i have adb/fastboot flashed boot/recovery/system, ran both zip files, after reboot, wont boot into ubuntu [11:19] supposed to be easy as its a nexus 4, supported device === ara is now known as Guest11257 [11:20] Guest21864, please use ubuntu-device-flash, manual methods have not been tested in months and most likely are not working [11:22] i am not good with ubuntu desktop, i need to go by manual methods otherwise i complicate myself lot more to get ubuntu-device-flash working [11:22] i did it on a nexus 10, nexus 4 just does not want to boot [11:23] does secure boot - enabled + lock state - unlocked look ok ? [11:23] if you could flash already all should be as unlocked as you need it [11:26] yes i was able to flash, boot into recovery, see ubuntu logo, send first zip by siteload, send second zip to siteload -> all ok saw progress bar that it was installing, after finish just didnt boot [11:30] * ogra_ goes and deletes these zips from the server ... they are months old [11:33] ogra_: can i write you in private ? === mandel is now known as mandel|lunch === alan_g is now known as alan_g|lunch === jhodapp|afk is now known as jhodapp [12:51] oSoMoN: I'm cleaning the landing-013 now that it got to release pocket, so that you'll get the other webbrowser landing started [12:54] Mirv, excellent, thanks! === _salem is now known as salem_ === dandrader is now known as dandrader|afk === alan_g|lunch is now known as alan_g [13:05] dpm: ping [13:06] anyone know why there is no roomservice.xml or even a local_manifests folder in my workspace? :D [13:06] a what? [13:09] in the porting tut on the ubuntu touch website it says that you have to edit the roomservice.xml in the local_manifests folder in order to get things working [13:09] but somehow there is nothing like that [13:09] oh [13:11] hi dobey [13:11] so, anyone know why? :D [13:17] Electropug, i dont think AOSP uses that [13:17] then why is it in the official tutorial :o [13:17] because we all suck at keeping the docs up to date :( [13:18] dpm: hi. for the .desktop files of click packages, are the translations being stripped out? and if so, why? we aren't building translation packages for all the clicks, so i don't see a need to strip them [13:19] dobey, there is nothing that strips translations of .click packages, regardless of them being .desktop or app translations. The .mo files are always shipped in the .click package [13:19] dpm: ok, great. [13:19] so, what can i do now ogra? :D [13:20] dpm: do you know if the manifest for the click has translations too? [13:20] there was some discussion recently on the mailing list ... take a look there [13:21] dobey, it hasn't. Translations for the manifest (i.e. the description) are done in the app store, although I've yet have to test it. [13:21] where do I find it? :) [13:23] dpm: oh ok. i thought mmaybe it was pullig them from the click manifest [13:23] dobey, that'd been my preference, so that translations also work when offline, but they come from the store [13:24] how does the store get them if they aren't in the manifest? [13:25] ogra_ , where do I find the mailing list? :) [13:27] are any of the core apps translated already? [13:27] dobey, you can upload a text file with translations in the store [13:28] and yes, all core apps are translated [13:28] the description in the store too? [13:28] the weather app desktop file seems to have translations stripped (has the x-ubuntu-gettext-domain instead) [13:29] no, we've not done the manifest translations for core apps in the store, that's what I meant by I haven't tested it yet :) [13:29] are they translated in the manifests? [13:30] dobey, the x-ubuntu-gettext domain is for the translation to be loaded from the .mo file instead of inline translations, it's got nothing to do with stripping (IIRC) [13:30] the local manifests are not translated, no, there is no specific format for translations in there, or to load them locally [13:30] dpm: how is that useful though, given clicks don't have language packs? [13:31] shouldn't the .desktop file have the translations inline instead? [13:31] for click packages translations are not installed in the regular /usr/share locations as in language packs [13:31] dpm: or is it just doing it this way because it's not using intltool? [13:32] thus click translations are independent from langpacks [13:32] yes i know that [13:32] I have LG p500, can I compile ubuntu touch code for my phone? [13:32] what i mean is, without langpacks, there seems to be no usefuless to the domain rather than having the translations in the .desktop file [13:32] ? [13:33] oh, I see what you mean now [13:34] the advantage then is probably simply the fact that we then don't need an extra cmake rule to merge the inline translations into the .desktop file [13:34] dpm: right, so it's just because you can't use intltool, so you'd have to do it manually? [13:35] how can I take a screenshot on the phone? [13:35] dobey, essentially, yes [13:35] ok [13:35] * dpm tries to think if there are other benefits from having everything in .mo files [13:36] I think that's mostly it, then [13:36] yeah. i don't think the amount of bytes saved from not having them in the .desktop is enough to call it an advantage really === mandel|lunch is now known as mandel [13:42] Can I compile Ubuntu touch code for LG optimus one(P500)? it has 512 MB RAM and 8GB memory [13:43] grepped, i guess no, it just support Nexus [13:44] ohh, thanks Mihir..but what do you mean by "it just support Nexus". For sure it supports tonnes of devices other than nexus [13:46] grepped, https://wiki.ubuntu.com/Touch/Install#Supported_devices_and_codenames [13:46] list of supported devices [13:49] grepped: there are unofficial ports to other devices [13:49] Problem with boot after Ubuntu touch | http://askubuntu.com/q/442872 [13:49] you are free to try to port it to your device [13:49] thanks dobey [13:50] is there any wiki/guide to port on my device? How could I know minimum hardware requirements to run ubuntu touch [13:52] grepped, link is in the topic === ara_ is now known as ara [13:53] mihir, pmcgowan, dobey ..thanks. I got the link. === dandrader|afk is now known as dandrader [14:10] dpm: is it possible to use i18n.tr from c++? if so, how? [14:11] dobey, according to bug 1182577 it is, but I've not tried it yet [14:11] bug 1182577 in Ubuntu UI Toolkit "[i18n] Cannot use the i18n plugin in C++" [Wishlist,Fix committed] https://launchpad.net/bugs/1182577 [14:11] looking at the branch, I think it should be Fix Released, though [14:12] so in theory UbuntuI18n::tr("Translate me!"); should work [14:12] but kalikiana can probably give you better advice [14:12] dpm: there isn't a c++ library is there? [14:13] kalikiana: ^ do we have a lib for that? or should the project link to the qml plugin?? [14:13] kalikiana: @ https://bugs.launchpad.net/ubuntu-ui-toolkit/+bug/1182577 [14:13] Ubuntu bug 1182577 in Ubuntu UI Toolkit "[i18n] Cannot use the i18n plugin in C++" [Wishlist,Fix committed] [14:13] dobey, I'd ask the sdk guys how to use it, I'm not an expert on this one. All I know is that the i18n plugin is written in C++ instead of QML, so it might be usable from C++ [14:14] but I see t1mp already stepped in :) [14:14] dpm: right, because it provides a qml component [14:14] in any case, this will need to be resolved to ensure scopes are translatable (bug 1297889) [14:14] bug 1297889 in unity-scope-click (Ubuntu) "Add i18n support to scopes" [High,Triaged] https://launchpad.net/bugs/1297889 [14:15] even if it is usable from c++, i'm not sure if we can do that in a scope (because it's not an app that has the qml runtime or any gui loaded) [14:15] maybe i need to just use raw gettext or glib gettext api [14:15] dobey: UbuntuI18n* i18n = &UbuntuI18n::instance();; i18n->tr(QString("Foo")) [14:15] kalikiana: where does UbuntuI18n come from? [14:16] ui toolkit [14:16] technically speaking you can also directly use gettext [14:16] how does one use that in c++? [14:16] that's what this is above :-) [14:16] no [14:16] where does that API come from? what header file? === brendand is now known as brendand-afk [14:17] i18n.h [14:18] I don't know if there's a nice example of it, few people use c++ for gui [14:18] yeah i'm not doing gui [14:18] i'm trying to add i18n support to the scope [14:25] bfiller: hi! renato tells me that you need help in debugging OA [14:25] bfiller: the first thing is uncomment the "LoggingLevel=2" line in /etc/signond.conf [14:25] bfiller: and kill signond, if it's running [14:26] bfiller: if the reason of the failure is that a web authentication is failing, you can debug signon-ui: [14:26] export SSOUI_LOGGING_LEVEL=2 [14:26] export SSOUI_DAEMON_TIMEOUT=9000 [14:27] bfiller: then kill signon-ui, and start it from the command line [14:28] dobey: does the scope have a MainView? I don't really know much about those… [14:30] kalikiana: no, it doesn't have gui. the gui is all in unity8. it just sends messages to the dash [14:31] mardy: ping [14:33] dednick: hi! [14:34] mardy: hi :) just wanted to follow up on some things with trust sessions [14:36] mardy: so, we should have ability to add by app id soonish. Are you waiting for this, or have you been working around it? [14:37] didrocks: so most likely we will not get a promoted image today, correct? [14:38] dednick: I've been waiting, since the way trust session work can have a huge impact on Online Accounts [14:39] dobey: maybe then just _("bla") is all you want, if you use nothing else from qml/toolkit [14:40] dednick: if a process with that app-id is already running, it won't be affected, right? [14:40] kalikiana: yeah, looks like i18n::tr isn't usable here [14:41] mardy: um, at the moment it will be added as part of the trust session. [14:42] dednick: OK; which means that you don't support having multiple instances of that [14:43] mardy: no. how would we distinguish between instances? [14:43] kenvandine: are the content hub designs published somewhere? [14:43] dednick: eh, I'm here for bringing trouble, not answers ;-) [14:43] dednick: did you see my last comment to your MP? I wrote a suggestion there [14:44] mardy, in google docs [14:44] one sec [14:44] cwayne: yeah, likely not [14:46] mardy: ah, no. i havent seen it yet [14:47] mardy, https://docs.google.com/a/canonical.com/document/d/1trse15NokU8IJ5lm3BnUi7oMNTCkUnYNHeAHZdtzFoQ/edit [14:51] kenvandine: I can't find the answer for this question, but maybe you can help: [14:51] kenvandine: if app A wants to import content from app B, and B is already running, what should happen? [14:52] kenvandine: would a new instance of B be started? [14:52] focus switches to B and B changes it's view to a picking view [14:52] however... that is today [14:52] when we have trust sessions [14:52] and multi-instance [14:52] we'll start a new instance [14:52] kenvandine: yep, I care about the future. OK, thanks [14:53] np [14:53] dednick: so, eventually we'll need multi-instance support also for the content hub ^ [14:53] dednick: it's fine if the first implementation leaves that out, but I think that at least you should take it into account in the API [14:54] dobey, which particular scope are you trying to internationalize? [14:55] hello guys, can anyone tell me why there is no roomservice.xml or local_manifests folder in my workspace like it should be? :) [14:57] mzanetti: for the time being, I believe there will only be one trust session running at a time. [14:57] mardy: ^ [14:57] mzanetti: sorry :) [14:59] mardy: well, to be fair, if you need truely multi-instance support, you need to use a unique identifier. For the time being we could probably assume the "next" session that opens with that name is the one to use if we give a name. [14:59] dednick: yep [15:02] dpm: click [15:04] cjwatson, So when click creates the directory to put links in for user hooks, what permissions does it use for that directory? [15:05] given that you're asking, I assume an answer is "the wrong ones" [15:05] cjwatson, Heh, I think so, but I could be wrong :-) [15:06] cjwatson, We're getting a url-dispatcher directory created with 600, and I'm trying to figure who's doing it. [15:06] it's supposed to be 0777 & ~umask [15:06] cjwatson, I think it's the click user hook putting those in. [15:06] it just does ensuredir, which does DirUtils.create_with_parents (directory, 0777) [15:06] mardy: I'll give that try thanks [15:07] so I'm prepared to believe you that I might be doing something wrong but I don't immediately see where [15:07] cjwatson, I'm looking at bug 1290997 [15:07] bug 1290997 in url-dispatcher (Ubuntu) "click crashed with gi._glib.GError in run(): Child process exited with code 139" [High,Triaged] https://launchpad.net/bugs/1290997 [15:07] cjwatson, Hmm, okay, I was more thinking perhaps you were doing the 600 on purpose. [15:10] tedg: I don't *think* so [15:10] if I am it's absolutely a mistake [15:10] dobey, ah, cool. So on the question of .desktop file translations, the old click scope did not load them from the .mo files, so all app names appeared in English. While you're looking at how to use translations from C++, could the new scope do that (i.e. display the .desktop translations? === brendand-afk is now known as brendand [15:16] pmcgowan: Did you happen to hear anything about the work on the alarm ringing during sleep? [15:16] nik90, not this week, last I heard the MR was waiting for some tweaks for desktop [15:17] rsalveti, might know more ^^ [15:17] pmcgowan: ah ok [15:20] hi there [15:20] I'm using the daily build of 14.04, fully updated AFAIK === Hi is now known as Guest14516 [15:21] seems that the QtGui suffers something which is very similar to https://bugreports.qt-project.org/browse/QTBUG-32760 -- when I ssh into the VM with SSH forwarding and launch a Qt5 app done in QML [15:22] the keyborad doesn't really type anything into the text field, and here's what I get in the console, http://fpaste.org/91320/39653507/ [15:22] I'm trying to find out how is your version of QtGui built, whether it uses the system's own libxkbcommon, or the Qt-bundled one [15:23] seems that this is actually https://bugs.freedesktop.org/show_bug.cgi?id=68056 , so you might either want to upgrade your libxkbcommon, or the version bundled with QtGui, depending on what you use [15:23] Freedesktop bug 68056 in General "Fails to compile czech(qwerty) keyboard" [Normal,Resolved: fixed] [15:24] popey mhall119^^ who is best to help with this? [15:25] jkt: Mirv may be able to help with that? [15:26] jkt: is that installed inside a vm or on real hardware? [15:27] Why wouldn't sound be heard on the phone when it works on the desktop with an Ubuntu Touch app? | http://askubuntu.com/q/442918 [15:28] popey: vm, libvirt, rhel6 [15:28] popey: connecting from a Gentoo machine, though, so the rhel6 should have no impact on this [15:28] jkt: and it's a clean install of ubuntu 14.04 then updated? [15:28] popey: installed yesterday, updated an hour ago [15:28] wow, that's quite a lengthy bug report [15:29] let me see if I can reproduce that here [15:29] bzoltan: are you available today to help me? [15:30] all right, this is because the upstream (libxkbcommon) commit fdb4de1f85e336be01f077e8c8e86c6a76bab6ba is only included in xkbcommon-0.4.1 [15:30] you guys use 0.3.1-2 [15:30] interestingly I get a completely different error when I try this. [15:31] popey: my primary KB is en_us, the second one is cz(qwerty) [15:31] $ xprop -root | grep -i xkb [15:31] _XKB_RULES_NAMES(STRING) = "evdev", "pc105", "us,cz", ",qwerty", "caps:internal,grp:alt_shift_toggle,grp_led:scroll,altwin:super_win,terminate:ctrl_alt_bksp" [15:33] ah, i wouldn't see that, only one layout [15:33] i forsee other issues though [15:33] messing with VNC is something that I would love to avoid [15:33] jkt: http://paste.ubuntu.com/7199196 [15:34] it'll need GL, which you wont get over X forwarding? [15:34] popey: try ssh -Y [15:34] popey: works fine here [15:34] cjwatson, If I install a new package that has click hooks, and that user is logged in, are the user click hooks run? [15:34] jkt: same [15:34] popey: but it seems to work just fine with -X here as well, strange [15:35] hmm [15:35] tvoss: sergiusens mentioned that h/w video encoder are to be enabled through media-hub [15:35] tvoss: do you have details on that? [15:35] popey: I've enabled http://ppa.launchpad.net/ubuntu-sdk-team/ppa/ubuntu, forgot about that [15:35] jkt: I'm trying to run the build dholbach did in the ppa. [15:35] that should be fine. [15:35] popey: I'm running what I built locally [15:36] popey, which build? [15:36] popey: which is what is in git, and I frankly don't care about what other have built :) [15:36] dholbach: one from your ppa [15:36] heh [15:36] popey, are we talking about trojitá? [15:36] yes [15:36] jkt: I'll build locally and see then [15:36] tedg: Installations are generally done *by* a user, and installations involve a "registration" step which basically makes that package available in the calling user's namespace; registration runs user-level hooks [15:37] popey, ah ok... are you talking about the armhf build which fails? [15:37] no [15:37] thanks [15:37] tedg: The case where installations aren't done by a logged-in user are preinstallations, and the Upstart job to run user-level hooks on login is intended mainly to cope with that situation [15:37] cjwatson, I mean with dpkg, so installation of a new url-dispatcher [15:38] jkt, I was just taking a look at some of the build logs of the git import of trojitá and noticed that the armhf builds failed (after running for 4+ hours) - which information, apart from the build log, would be useful? [15:38] cjwatson, So what does dh_click do, I guess. [15:38] Does grouper work at all these days? [15:38] can anyone here help me? :D [15:38] tedg: That doesn't run user-level hooks, no [15:39] tedg: It arguably could, but I'm cautious of anything that involves having to enumerate users [15:39] cjwatson, And running user level hooks is synonymous with setting up the link farm for user level hooks, right? [15:39] dholbach: the build log is a good first step [15:39] tedg: Not synonymous exactly, but the latter is part of the former. [15:39] jkt, all right, thanks - I'll send in a bug report in a bit then [15:40] (Hooks can also define a command to execute to catch themselves up with the state of the link farm.) [15:40] dholbach: oh, and make sure you build with -DWITH_RAGEL=off if you're cross-compiling [15:40] K, trying to figure out who's making this directory. Nothing obvious :-/ [15:40] dholbach: there's a bug in upstream ragel (reported half a year ago...) where it assumes that host's char is the target's arch, which is horribly wrong on arm [15:40] tedg: Do we have a reproduction recipe yet/ [15:40] ? [15:40] aha! ok, do you have a link to the bug in question? [15:41] cjwatson, No, it seems that once people fix it, it goes away. [15:41] tedg: If so I'd be tempted to suggest just stracing the lot ... [15:41] Ah :-/ [15:41] I don't think I've upgraded to the new url-dispatcher yet. I wonder if it will happen to me [15:41] dholbach: but I doubt you've installed ragel, it's a pretty exotic package [15:41] can anyone tell me why there is no roomservice.xml or local_manifests folder in my workspace like it should be? :) [15:41] Before you do check to see if you have the dir already. [15:41] Oh, I don't even have it installed. [15:41] Which directory again? [15:41] I'm curious if it got created previously incorrectly. [15:41] ~/.cache/url-dispatcher [15:42] ~/.cache/url-dispatcher [15:42] * Laney wins [15:42] Again [15:42] dholbach: http://www.complang.org/pipermail/ragel-users/2013-September/002989.html [15:42] Right, doesn't exist here. [15:42] jkt, thanks [15:42] So, I expect you won't have problems. But it'd be interesting to see. [15:42] That'd test the legacy theory :-) [15:43] dholbach: and https://www.mail-archive.com/ragel-users@complang.org/msg00929.html for the full thread [15:43] jkt, ok, turns out the build was killed because it took ages - the only thing I could spot (when running it locally through qemu) was 2 tests failing [15:43] dpm: displaying the translations for the app names/descriptions themselves is a separate and harder problem. i think we really need to get away from having the translations in the .mo files for that [15:43] jkt, I'll file a bug about the failing tests - and will find out what we can do to allow some of the builds taking a bit longer in PPAs [15:44] jkt, http://paste.ubuntu.com/7199236/ is what happened [15:44] dholbach: use xvfb-run for tests, they need X [15:45] dholbach: see the debian stuff in packaging/obs-trojita*/... for a working debian/rules anyway [15:45] dholbach: also, please run tests with `ctest --output-on-failure` to get a meaningful output [15:46] dobey, I'm not talking about the descriptions on an app's individual page. I mean the dash page that shows all installed apps. Could the translations not be loaded from the .desktop files -regardless of whether we keep them in the .mo files or inline-, at least in offline mode? [15:46] jkt, hum... the build worked just fine on i386 and amd64 and the packaging is very much influenced by what's in ./packaging/ :-) [15:46] jkt, thanks - I'll have a look at "ctest --output-on-failure" [15:48] AlbertA, best to talk to jhodapp [15:53] AlbertA, yes that's the plan, it's my next major task to work on [15:53] popey: after building libxkbcommon from git, my keyboard now works [15:53] hurrah jkt [15:53] popey: you might want to make sure that you use a recent version of that lib [15:54] most .cz people would typically use a qwertz layout, but still, this is a pretty annoying bug [15:54] jhodapp: ok, I have these changes to enable h/w encoding from a buffer using stagefright [15:54] jhodapp: https://code-review.phablet.ubuntu.com/#/q/owner:%22Alberto+Aguirre+%253Calberto.aguirre%2540canonical.com%253E%22,n,z [15:55] jhodapp: just wondering if it would fit with the media-hub plans...my plan was to use it so we actually [15:55] jhodapp: record more than a couple of seconds with mirscreencast [15:55] jhodapp: on the device [15:55] jkt: as a temp fix (for our users) we could build a newer version of libxkbcommon in our ppa while we get the archive version updated, dholbach sound plausible? [15:56] AlbertA, I'll look over your changes and let you know [15:56] AlbertA, hopefully they are usable [15:56] popey, do we have a bug report for libxkbcommon or something - I don't know which other effects it has? is this a problem on the dekstop? on the phone? [15:56] rsalveti: can you elaborate the status of the alarm ringing while the phone is in deep sleep please? I need that for the meeting today. [15:56] dholbach: not yet, desktop [15:56] nik90: rsalveti is away for the day [15:56] dholbach: i only learned of it from jkt just now. [15:57] nik90: i know the status ☻ [15:57] jhodapp: ah..thnx. I will catch him tomorrow [15:57] i think he is out all week [15:57] popey: oh ok :) [15:57] nik90: he's off for this week [15:57] dholbach: https://bugs.freedesktop.org/show_bug.cgi?id=68056 , https://bugreports.qt-project.org/browse/QTCREATORBUG-9978 , https://bugreports.qt-project.org/browse/QTBUG-32760 [15:57] Freedesktop bug 68056 in General "Fails to compile czech(qwerty) keyboard" [Normal,Resolved: fixed] [15:57] jhodapp: ok [15:57] dholbach: tag 0.4.1 is OK, 0.4.0 doesn't contain that fix, unfortunately [15:58] jkt: hmm, built trojita from git and I still get the libGL issue. [15:58] surprised you don't. [15:58] tedg: not a legacy problem, it fails for me [15:58] jkt, I'll talk to the guys in #ubuntu-x about it [15:58] but I wont worry about it because I will probably never run it over ssh -X, and worry less if it works for you. [15:59] $ ls -ld .cache/url-dispatcher [15:59] drw------- 2 cjwatson cjwatson 4096 Apr 3 16:42 .cache/url-dispatcher [15:59] thanks dholbach [15:59] Oh! great data point. [15:59] cjwatson, Did you have to logout and back in, or just on install? [15:59] dholbach: you might want to ask the Qt guys about that library; my impression is that given it's a pretty new stuff and AFAIK nobody but Qt5 uses it, the old releases might be rather buggy [15:59] dholbach: ok [15:59] tedg: Repeatable by purging url-dispatcher, removing that directory, and reinstalling url-dispatcher. [15:59] tedg: Just on install. [16:00] tedg: Let me strace it. [16:00] Oh, my. [16:00] That means we're doing something per-user on install. [16:00] jkt, ok, tanks [16:01] tedg: click might be walking over the active user registrations and attempting to update them [16:02] Though that seems a bit weird. [16:02] Hm, I tried that and it had the correct permissions after reinstallation [16:03] tedg: Wait, this is bizarre, I don't see the mkdir in the trace ... [16:05] Could installing the user session jobs be kicking one off? [16:05] I was just wondering that [16:06] tedg: Stopping all the url-dispatcher user jobs, then "rm -rf .cache/url-dispatcher; start url-dispatcher" does it for me. [16:07] tedg: I think this rules out click ... [16:08] cjwatson, Okay, I was able to get that to happen. [16:08] cjwatson, Thanks, will look into it further. [16:10] I don't know if this is related, but why aren't you using the standard library's S_IRWXU constant from rather than writing "g_mkdir_with_parents(urldispatchercachedir, 1 << 6 | 1 << 7 | 1 << 8); // 700"? [16:10] It'd be much clearer than bit-shifting that you have to comment ... [16:10] Didn't think of it at the time... [16:10] No good reason. [16:11] Oh, it doesn't happen if I just run the dispatcher. Only under upstart. [16:12] I had to check that the relative precedence of << and | was that way round, but it does seem to be [16:14] :-( strace has it as 0700 [16:15] Yeah, I was just going to say [16:15] Mad umask maybe? === _salem is now known as salem_ [16:16] tedg: the umask is apparently 0117 [16:16] this is a bit insane [16:16] Where does that come from? [16:17] A fine question [16:21] So indeed setting the umask to 0002 in the upstart job fixes it. [16:21] Mm. I think we should figure out where this comes from though. [16:21] Not sure if that's a fix or a work around [16:23] creating a new /usr/share/upstart/sessions/test.conf that's just description "test" / manual / exec sh -c umask and then doing "start test" results in 0117 in .cache/upstart/test.log [16:24] init's own umask is also 0117 [16:24] so my guess is that that was its initial umask when started === jhodapp is now known as jhodapp|lunch [16:26] Xsession.d ? [16:27] $ grep umask /etc/X11/Xsession.d/* [16:27] $ [16:27] There's a check in /etc/X11/Xsession but no setting. [16:27] And all of the STARTUP accumulation stuff is discarded if you're using an upstart session [16:28] Xsession sets it, but only in a subshell [16:28] Let's see if a guest account does it [16:30] Guest session does *not* do it; nor does my son's account [16:30] But my session has been up since Feb19 [16:31] I rebooted this morning… [16:31] I can't reproduce either, fwiw [16:32] Nothing relevant-looking in trusty-changes [16:32] I need to run for a bit. I can keep looking when I get back. Weird bug. [16:34] popey, see the discussion with tjaalton in #ubuntu-x [16:34] * popey looks [16:34] lool: ping -- re ubuntu-touch-customization-hooks [16:35] tedg,Laney: Ha, well, so much for that - I tried to gdb lightdm to find out what its umask was but I confused it horribly and it crashed my session [16:35] tedg,Laney: So now I can't reproduce it any more [16:36] tedg: I think you're probably best with defensive measures to make sure you have a sane umask, maybe fix the permissions on the directory if they're broken, and leave it at that [16:36] hah [16:38] kenvandine: ping [16:40] tedg: ping [16:40] dednick, pong [16:41] cjwatson: Got it [16:41] kenvandine: howdy. was it you I was talking to trust sessions last week? [16:41] It happens if you re-exec upstart [16:41] dednick, nope [16:41] Oh, is it not serialising it properly? [16:41] dednick, but i am waiting to have them :) [16:41] popey, bfiller, can you tell me which of the list of apps from the whiteboard you own? https://blueprints.launchpad.net/unity8/+spec/suru-icon-switch [16:41]